Attiny 417 output failure

nobner · Joined: 29 Apr 2015 Posts: 21

Hello to all,
i'm using a Attiny 417 for driving RGBW-LED.
4x AD-Inputs, 8 bit, for checking dimming range.
4x Digital Outputs with variable Puls-Length for driving LED power supply.
3 Channels are working fine, the fourth channel (PortB.1) is producing double pulses (see attached pictures)
Program is the same for all 4 Channels.
I think i had a similar problem years ago and someone told me, that there is an alternative to set the output-ports (i'm using PortB.1 = 1)
Unfortunatley i can't find the post anymore.
Any idea's what the problem could be?
Are 4 AD-Inputs too much for stack ,or....?
Has there anything to be disabled for using PB.1 as digital output?

Thank's !!

(BASCOM-AVR version : 2.0.8.4 )

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

if a pin does not work always start with just toggling the pin.
also, trace/measure with a DVM on your PCB to see if that is the actual pin.

then check the 5.1 section in the PDF. it shows the alternative pin functions.
AIN10, XCK, SDA, WO1

thus when using TCA0 with WO output you could get into problems.

there is builtin support for rgbw :leds config rainbow

to set a pin to output you can best use CONFIG PORTB.1=output
SET PORTB.1 and PORTB.1=1 are equivalent

of course you can use the DDR and PORT register too but this makes code less portable.

nobner · Joined: 29 Apr 2015 Posts: 21

Hi Mark,
thank you for the hints.
I did measurement Input / Output. All 4 Channels are showing the same (correct) input signal and 3 channels are doing the correct output.
Only channel 4 (PB1) is giving the failure output.
How can i disable the alternative pin functions for AIN10, XCK, SDA, WO1?
Writing registers on xtiny seem to be different to the standard tiny's?

UPDATE:
one step further to solution:
When i disable the input section of one channel (here the white one) in software, PORTB.1 works fine.
Might this be a stack problem?

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

that does not make sense. i would start with a verify and see if the code is programmed.

you hardly use stack. and all the color sections are equivalent.

but pb1 is also AIN10. so when using the A/D. you best put the CONFIG PORTB.1 after the config adc0

i could not find explicit info on a/d input pins digital mode usage.

nobner · Joined: 29 Apr 2015 Posts: 21

Changing config Portb.1 behind config adc doesn't change anything.
Programming work's without problem - three channels working correct, the fourth, PB1 not.
I don't use the AD-input pins in digital mode? What do you mean with this comment?
I wonder why the frequency of the failure output is exactly 4 times higher than the correct one?

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

nobner · Joined: 29 Apr 2015 Posts: 21

I understand what you mean, but as i know from fewer projects with e.g. Atmega328 there was never a problem to use a ADC Pin as an input and another pin as an digital output, thus this was an possible ADC Pin also. It even was not necessary to stop the ADC for that.
One post before i asked you about section 5.1 in the datasheet - how can i disable the alternative functions? With Atmega i used register writing to do so, but with xtiny it seems different? How can i disable the alternative functions? As "Stop ADC0" is not working, how can i stop the ADC during program execution?

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

Config Adc0 = Single , Adc = disabled

or wrote 0 to adc0_CTRLA.0 (register names and locations vary from chip to chip) so best use the config as shown.

i still did not see a reference to the used hardware.

nobner · Joined: 29 Apr 2015 Posts: 21

Hi Mark,
as suggested i tried to disable Adc after use - no effect on the failure.
My momentary failure conclusion: Using 3 Adc inputs - is working, using 4 Adc inputs - is not working. Can there be a kind of Adc-overflow?
Attached my fuses and the circuit diagram.

Norbert

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

it is not a stack problem. stack comes when you call sub routines. but you do not have them. you also do not use interrupts.
a/d overflow is not likely.

you can best add some print statements and print the values you measure. and see if all code is executed with values you think they should have.
PortA.1 can server as TX pin.
to get the alternative tx pin, use :
config COM1=9600, TXPIN=ALT1_PA1
or
Config Port_mux = Preserve , Usart0 = Alt1_pa1pa4

you can also try a modification on your pcb and use another pin for the output.

it will also be a good idea to check the vcc with a scope. also i do not know if there is a load on the circuit but try to measure the output without load.

then you can also try to fake the a/d value. just give it some hard coded value.

MWS · Joined: 22 Aug 2009 Posts: 2262

With a certain understanding to the beginners worries and problems, I would suggest something like that:
1) open trash bin
2) place code in trash bin
3) close trash bin
4) select empty trash bin

While it still may make sense, to execute on-time of different colors "one after each other", it reduces maximal brightness to one quart at default current.
The question is: why would one do or want that?

I can't also make sense out of the fact, that any ADC input value over 240 shortens runtime of the main loop and thus frequency of the PWM.
If the TO intends this behavior I'd surely am interested to hear about the reason.

Even if one does not want or does not know how to use hardware PWM or interrupt-driven soft-PWM, one can very easily do a soft-PWM in the main loop, which wins against this botchery by lengths.

If then a simplified and logically improved code exists, looking for errors is easier too.

albertsm · Joined: 09 Apr 2004 Posts: 5921 Location: Holland

MWS, it does not resolve the portb.1 problem.

MWS · Joined: 22 Aug 2009 Posts: 2262

MWS · Joined: 22 Aug 2009 Posts: 2262

Thought about something like the following, may have errors, have no xtiny-libs.
Calculated 4 channels two times 13 ADC-ticks at 2.5MHz ADC-clock to 52µs, which results in a PWM repeat rate of 75Hz.
It will perform differently than the other code, as getting the analog values is part of the PWM-loop.
The correct way is to use HW-PWM, the 417 seems to sport 6 PWM channels.

nobner · Joined: 29 Apr 2015 Posts: 21

Hi Mark,
finally i found the solution:
It is the name of the variable "RED_PAU". When i replace it with some other name (like "nonsense", or any other name), programm works bug-free.
Why is this so?
Thanks for support.

WORKS NOT: